Thread: difference in output power measurement


Permlink Replies: 9 - Pages: 1 - Last Post: Jun 26, 2012 3:46 PM Last Post By: daras Threads: [ Previous | Next ]
nsm

Posts: 39
Registered: 02/20/12
difference in output power measurement
Posted: Jun 18, 2012 8:05 AM
Click to report abuse...   Click to reply to this thread Reply
Hi,

I am trying to characterize the output power of a high power device(peak o/p power 46dBm) using a network analyzer and also a power meter. I am doing the measurements in the pulsed mode.

The output power and the insertion gain is measured at -13dBm.

1. I have done a source power cal and an enhanced response cal for measuring the S21 and the S11 measurements. I get a gain of approx 68dB. The input return loss is at -4dB(very poor).
2.For doing the output power measurements on the DUT , I do a source power cal on the network analyzer at the frequencies I need to measure the output power.
3. I find that there is a difference of more than a dB between the power indicated on the power meter and the o/p power computed from S21-input power.
4. I am using a total attenuation of 40dB in the path. The power meter sees a good match at the output of the device.

Is the difference in the power calculated and measured because of the poor return loss. The DUT is also in compression.

Could you please clarify.

Rgds
nsm
daras


Posts: 832
Registered: 12/01/05
Re: difference in output power measurement
Posted: Jun 18, 2012 8:20 PM   in response to: nsm in response to: nsm
Click to report abuse...   Click to reply to this thread Reply
So on your VNA you must be measuring S21 under pulsed condition (i.e synchronized to measure only during pulse on), but are you also measuring with a triggered peak power sensor? if not then the power sensor is measuring the average power of the pulse and the reading would have to be corrected for the pulse duty cycle. You didn't say what the duty cycle of the amp was and which way the 1 dB difference went, so it is hard to tell if the difference you are seeing can be attributed to the average versus peak power measurement methods.
nsm

Posts: 39
Registered: 02/20/12
Re: difference in output power measurement
Posted: Jun 19, 2012 12:01 AM   in response to: daras in response to: daras
Click to report abuse...   Click to reply to this thread Reply
Hi Daras,

we are doing the measurement with a duty cycle of 5%. We measure the S21 under the pulse condition with an enhanced response cal.

for the peak power measurement we are using the N1911A power meter with the N1921A peak power sensor. For the peak power measurement, we set the NA in the CW mode after performing a source power cal. I find that the power meter shows a lower value as compared to the power computed from the network analyzer gain measurement.

Rgds
Nsm
daras


Posts: 832
Registered: 12/01/05
Re: difference in output power measurement
Posted: Jun 19, 2012 6:40 PM   in response to: nsm in response to: nsm
Click to report abuse...   Click to reply to this thread Reply
What model network analyzer are using (if it is a PNA, pleas also indicate the firmware rev)? Are you measuring the input power with the network analyzer reference receiver or are you assuming that with the source power cal, the input power is going to be exact and flat? You said that the input return loss is very poor, so that usually means that the power delivered to the DUT may not be exactly the power that you measured with a power sensor during the source power cal.
nsm

Posts: 39
Registered: 02/20/12
Re: difference in output power measurement
Posted: Jun 19, 2012 7:45 PM   in response to: daras in response to: daras
Click to report abuse...   Click to reply to this thread Reply
hi daras,

I am using the E8362C NA(A.09.33.09). I am not using the reference receiver to measure the source power. I have assumed that with the source power cal, the power at the output is flat and I have confirmed this by measuring the o/p on the power meter.

Would you recommend that , I measure the source power at the time of measurement using the refernce receiver?

Rgds
Mahesh
daras


Posts: 832
Registered: 12/01/05
Re: difference in output power measurement
Posted: Jun 20, 2012 6:00 AM   in response to: nsm in response to: nsm
Click to report abuse...   Click to reply to this thread Reply
nsm wrote:
Would you recommend that , I measure the source power at the time of measurement using the refernce receiver?

Yes - Most power sensors have very good match and therefore when you measure the port power with a sensor with a very good return loss, most of the power from the source will be delivered to the sensor. On the other hand your DUT that has a poor return loss, reflects more of the power back into the source. A corrected S21 measurement accounts for this mismatch, but if you want to use the S21 to compute the output power, you can't assume a flat input incident power.

In order to measure the input power with your PNA, you will also need to do at least an R1 Receiver Power cal. However, for the most accurate measurement you will need a match-corrected power measurement. If you happen to have option 083 or 082, you can setup an Scalar Mixer measurement (SMC) with a 0 Hz offset and Simply use the OPwr measurement, which will be the match-corrected output power of your DUT.
nsm

Posts: 39
Registered: 02/20/12
Re: difference in output power measurement
Posted: Jun 20, 2012 9:58 AM   in response to: daras in response to: daras
Click to report abuse...   Click to reply to this thread Reply
Hi Daras,

The network anlayzer that I have does not have option 082 or 083.

I would like to clarify one additional point. I have done a source power cal and then an enhanced resposne cal for doing the S21 measurement. I would presume the source power would be flat to the extent of the tolerence specified during the source power cal.

To the above S21 measurement, I add the input power to compute the estimated o/p power.

Then i measure the output power of the device on a peak power meter by setting the network analyzer to CW sweep mode with a source power cal and using it as the source.

With this set up what could be the extent of the difference in the power meter readings and the value computed of the peak power from the measured S21? How can I estimate this?

Rgds
nsm

daras


Posts: 832
Registered: 12/01/05
Re: difference in output power measurement
Posted: Jun 25, 2012 11:16 AM   in response to: nsm in response to: nsm
Click to report abuse...   Click to reply to this thread Reply
I had to consult with Dr_joel to make sure I get this part right (and he may actually chime in here himself as he is a sucker for a good power discussion :-))

So let's start by looking at the equation for power delivered to a load (P del ) as a function of the incident power (P inc ) and S11 of the load:

P del = P inc * (1 - |S11|^2)


The built-in cal factors of the power sensors correct for this offset since the S11 of the power sensor is measured in the factory. So the readings produced by a power sensor are as if the sensor had a perfect 50 ohm match (which in many cases is not very far off). However, they also assume that the gamma of the source (source match) attached to the power sensor is also 50 ohms, which in the case of the source of a VNA is usually not very true. So in effect, the incident power changes as a product of the gamma source and sensor match:

P inc = P source_to_50ohm / (1 - Gamma source * S11)


So the accumulative effect of this is that the power delivered to and measured by a power sensor may not be the same as the power delivered to a DUT by the same source at the same set power level. the magnitude of the difference is related to the difference in the (1 - Gamma source * S11 psensor ) and (1 - Gamma source * S11 dut ). The match corrected power measurements in the PNA make this additional correction so that when input or output powers are measured the difference in source/load matches are accounted for.

A simple way for you to improve the agreement between the output power measured by a power sensor and output power computed by (S21 * InputPwr) is to put a 3-6 dB pad at the end of the cable connected to the input of the DUT and you should perform your source power cal and enhanced response cal at the end of that pad. The inclusion of the pad will reduce the difference in source/load match terms between the calibration and measurements.

Hope this answers your question.
nsm

Posts: 39
Registered: 02/20/12
Re: difference in output power measurement
Posted: Jun 26, 2012 11:06 AM   in response to: daras in response to: daras
Click to report abuse...   Click to reply to this thread Reply
Hi daras,

Thank you for the lucid explanation. The equations does help me account for the errors. However, the whole setup is a part of an automated test set.It would be difficult to add a pad at the input.Would there be any other method to do this measurement?

Rgds
mahesh
daras


Posts: 832
Registered: 12/01/05
Re: difference in output power measurement
Posted: Jun 26, 2012 3:46 PM   in response to: nsm in response to: nsm
Click to report abuse...   Click to reply to this thread Reply
nsm wrote:
Would there be any other method to do this measurement?

There is. In version A.09.33.09, which you are running, and in newer versions of the PNA firmware, there is a feature called "Guided Power Calibration". This feature allows you to augment a s-parameter calibration with a power meter cal (all part of the guided calibration wizard) and with that option, you not only get a s-parameter cal for the ports that you select, you also get a source power cal and a receiver power cal for all the measurement and reference receivers associated with the selected ports. so once you perform this calibration, in your channel you not only get corrected s-parameter measurements, you also get corrected receiver measurements such as R1 and B. These receiver measurements are actually match corrected power measurements. So, in your scenario (with an Enhanced Response Cal), the R1,1 trace would be the match corrected input power and B,1 would be the response corrected output power measurement. I did an example of this measurement on an amplifier with 30 dB gain to compare the measured B,1 power with an output power measurement made with a power sensor and also with an output power measurement calculated using (S21 * Input Power).

The first picture shows the steps involved in the cal:



The 2nd picture shows the results:



Note that Tr4 is a measurement made with a power sensor (using the PMAR feature) and Tr5 is an equation trace using the following equation:

Calc_Pout = Tr1 * Tr3


Given the model number of your PNA, you can actually upgrade your firmware to A.09.42.18, which I would highly recommend, as many improvements have been made in the newer version, especially in the area of the guided power calibration.

Point your RSS reader here for a feed of the latest messages in all forums